jinzd-ai-cli 0.4.54 → 0.4.56

package/dist/index.js

@@ -11,20 +11,23 @@ import {
   buildPhantomCorrectionMessage,
   buildWriteRoundReminder,
   clearDevState,
+  computeCost,
   detectsHallucinatedFileOp,
   extractWrittenFilePaths,
   findPhantomClaims,
+  formatCost,
   formatGitContextForPrompt,
   getContentText,
   getGitContext,
   getGitRoot,
+  getPricing,
   hadPreviousWriteToolCalls,
   loadDevState,
   parseSimpleYaml,
   saveDevState,
   sessionHasMeaningfulContent,
   setupProxy
-} from "./chunk-6FYFVPVE.js";
+} from "./chunk-2DC5ABAM.js";
 import {
   ToolExecutor,
   ToolRegistry,
@@ -38,7 +41,7 @@ import {
   spawnAgentContext,
   theme,
   undoStack
-} from "./chunk-TAR67QTH.js";
+} from "./chunk-FJSEFQ54.js";
 import {
   fileCheckpoints
 } from "./chunk-4BKXL7SM.js";
@@ -63,7 +66,7 @@ import {
   SKILLS_DIR_NAME,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-NP5KZVP6.js";
+} from "./chunk-7FOGK5TM.js";
 
 // src/index.ts
 import { program } from "commander";
@@ -487,8 +490,12 @@ Error${typeName}: ${lines.join("\n")}
   renderUsage(usage, sessionTotal) {
     const total = usage.inputTokens + usage.outputTokens;
     let line = theme.dim("\u{1F4CA} ") + theme.dim(`in ${usage.inputTokens.toLocaleString()}`) + theme.dim(" + ") + theme.dim(`out ${usage.outputTokens.toLocaleString()}`) + theme.dim(` = ${total.toLocaleString()} tokens`);
+    const cacheRead = usage.cacheReadTokens ?? 0;
+    if (cacheRead > 0) {
+      line += theme.dim(`  \u2502  cache: ${cacheRead.toLocaleString()}`);
+    }
     if (sessionTotal) {
-      const sessionSum = sessionTotal.inputTokens + sessionTotal.outputTokens;
+      const sessionSum = sessionTotal.inputTokens + sessionTotal.outputTokens + (sessionTotal.cacheCreationTokens ?? 0) + (sessionTotal.cacheReadTokens ?? 0);
       line += theme.dim(`  \u2502  session total: ${sessionSum.toLocaleString()}`);
     }
     process.stdout.write(line + "\n\n");
@@ -1217,11 +1224,18 @@ function createDefaultCommands() {
         if (sys) {
           console.log(`  System   : ${sys.slice(0, 60)}...`);
         }
-        const totalTokens = tokenUsage.inputTokens + tokenUsage.outputTokens;
+        const cacheRead = tokenUsage.cacheReadTokens ?? 0;
+        const cacheCreate = tokenUsage.cacheCreationTokens ?? 0;
+        const totalTokens = tokenUsage.inputTokens + tokenUsage.outputTokens + cacheRead + cacheCreate;
         if (totalTokens > 0) {
+          const cacheSuffix = cacheRead > 0 || cacheCreate > 0 ? ` [cache: +${cacheCreate.toLocaleString()} / -${cacheRead.toLocaleString()}]` : "";
           console.log(
-            `  Tokens   : in ${tokenUsage.inputTokens.toLocaleString()} + out ${tokenUsage.outputTokens.toLocaleString()} = ${totalTokens.toLocaleString()} (session total)`
+            `  Tokens   : in ${tokenUsage.inputTokens.toLocaleString()} + out ${tokenUsage.outputTokens.toLocaleString()} = ${totalTokens.toLocaleString()}${cacheSuffix}`
           );
+          const cost = computeCost(ctx.getCurrentProvider(), ctx.getCurrentModel(), tokenUsage);
+          if (cost != null) {
+            console.log(`  Cost     : ${formatCost(cost)} (session total)`);
+          }
         }
         const ctxWindowSize = ctx.getContextWindowSize();
         if (ctxWindowSize > 0) {
@@ -1823,7 +1837,7 @@ ${hint}` : "")
     },
     {
       name: "cost",
-      description: "Show session token usage summary",
+      description: "Show session token usage, prompt-cache hits, and USD cost",
       usage: "/cost [reset]",
       execute(args, ctx) {
         const sub = args[0]?.toLowerCase();
@@ -1832,24 +1846,65 @@ ${hint}` : "")
           ctx.renderer.printSuccess("Session token counters reset.");
           return;
         }
-        const usage = ctx.getSessionTokenUsage();
-        const totalTokens = usage.inputTokens + usage.outputTokens;
+        const session = ctx.sessions.current;
+        const usage = session?.tokenUsage ?? {
+          inputTokens: 0,
+          outputTokens: 0,
+          cacheCreationTokens: 0,
+          cacheReadTokens: 0
+        };
+        const cacheCreate = usage.cacheCreationTokens ?? 0;
+        const cacheRead = usage.cacheReadTokens ?? 0;
+        const totalTokens = usage.inputTokens + usage.outputTokens + cacheCreate + cacheRead;
         if (totalTokens === 0) {
           ctx.renderer.printInfo("No token usage recorded this session.");
           return;
         }
+        const provider = ctx.getCurrentProvider();
+        const model = ctx.getCurrentModel();
+        const cost = computeCost(provider, model, usage);
+        const pricing = getPricing(provider, model);
+        let savings = null;
+        if (cost != null && pricing && cacheRead > 0) {
+          const costWithoutCache = computeCost(provider, model, {
+            inputTokens: usage.inputTokens + cacheRead,
+            outputTokens: usage.outputTokens,
+            cacheCreationTokens: cacheCreate,
+            cacheReadTokens: 0
+          });
+          if (costWithoutCache != null) savings = costWithoutCache - cost;
+        }
         console.log();
-        console.log(theme.heading("  \u{1F4CA} Session Token Usage"));
-        console.log(theme.dim("  " + "\u2500".repeat(40)));
-        console.log(theme.dim("  Input tokens  : ") + chalk2.white(usage.inputTokens.toLocaleString()));
-        console.log(theme.dim("  Output tokens : ") + chalk2.white(usage.outputTokens.toLocaleString()));
-        console.log(theme.dim("  Total tokens  : ") + chalk2.bold.white(totalTokens.toLocaleString()));
-        console.log(theme.dim("  " + "\u2500".repeat(40)));
-        const session = ctx.sessions.current;
+        console.log(theme.heading("  \u{1F4B0} Session Cost & Token Usage"));
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
+        console.log(theme.dim("  Input (uncached) : ") + chalk2.white(usage.inputTokens.toLocaleString().padStart(12)));
+        console.log(theme.dim("  Output           : ") + chalk2.white(usage.outputTokens.toLocaleString().padStart(12)));
+        if (cacheCreate > 0) {
+          console.log(theme.dim("  Cache write      : ") + chalk2.yellow(cacheCreate.toLocaleString().padStart(12)));
+        }
+        if (cacheRead > 0) {
+          const pct = Math.round(cacheRead / (cacheRead + usage.inputTokens) * 100);
+          console.log(
+            theme.dim("  Cache read       : ") + chalk2.green(cacheRead.toLocaleString().padStart(12)) + theme.dim(`  (${pct}% hit rate)`)
+          );
+        }
+        console.log(theme.dim("  Total tokens     : ") + chalk2.bold.white(totalTokens.toLocaleString().padStart(12)));
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
+        if (cost != null) {
+          console.log(theme.dim("  Cost             : ") + chalk2.bold.cyan(formatCost(cost).padStart(12)));
+          if (savings != null && savings > 0) {
+            console.log(
+              theme.dim("  Cache savings    : ") + chalk2.green(`-${formatCost(savings)}`.padStart(12)) + theme.dim(`  (vs no cache)`)
+            );
+          }
+        } else {
+          console.log(theme.dim("  Cost             : ") + theme.dim("      \u2014 (pricing unknown)"));
+        }
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
         if (session) {
-          console.log(theme.dim("  Provider      : ") + theme.dim(ctx.getCurrentProvider()));
-          console.log(theme.dim("  Model         : ") + theme.dim(ctx.getCurrentModel()));
-          console.log(theme.dim("  Messages      : ") + theme.dim(String(session.messages.length)));
+          console.log(theme.dim("  Provider         : ") + theme.dim(provider));
+          console.log(theme.dim("  Model            : ") + theme.dim(model));
+          console.log(theme.dim("  Messages         : ") + theme.dim(String(session.messages.length)));
         }
         console.log();
       }
@@ -2106,7 +2161,7 @@ ${hint}` : "")
       usage: "/test [command|filter]",
       async execute(args, ctx) {
         try {
-          const { executeTests } = await import("./run-tests-P53FNUJY.js");
+          const { executeTests } = await import("./run-tests-GISOOQZC.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {
@@ -3282,7 +3337,7 @@ ${content}
 var FREE_ROUND_TOOLS = /* @__PURE__ */ new Set(["write_todos"]);
 var MAX_CONSECUTIVE_FREE_ROUNDS = 3;
 var MAX_REPEATED_TOOL_CALLS = 2;
-var AUTO_PAUSE_INTERVAL = 10;
+var DEFAULT_AUTO_PAUSE_INTERVAL = 50;
 function fmtTokens(n) {
   if (n >= 1e6) return `${Math.round(n / 1e5) / 10}M`;
   if (n >= 1e3) return `${Math.round(n / 1024)}K`;
@@ -3344,7 +3399,14 @@ var Repl = class {
   /** 当前加载的层级上下文（全局/项目/子目录） */
   contextLayers = [];
   /** 本次会话累计 token 用量 */
-  sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+  sessionTokenUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
+  /** Fold a single-request TokenUsage (with optional cache fields) into sessionTokenUsage. */
+  addSessionUsage(u) {
+    this.sessionTokenUsage.inputTokens += u.inputTokens;
+    this.sessionTokenUsage.outputTokens += u.outputTokens;
+    this.sessionTokenUsage.cacheCreationTokens += u.cacheCreationTokens ?? 0;
+    this.sessionTokenUsage.cacheReadTokens += u.cacheReadTokens ?? 0;
+  }
   /** 启动时检测到的 Git 分支（无 git 仓库时为 null） */
   gitBranch = null;
   /** MCP 多服务器管理器（无 MCP 配置时为 null） */
@@ -4607,8 +4669,7 @@ Session '${this.resumeSessionId}' not found.
         session.addMessage({ role: "assistant", content, timestamp: /* @__PURE__ */ new Date() });
         this.events.emit("message.after", { content });
         if (usage) {
-          this.sessionTokenUsage.inputTokens += usage.inputTokens;
-          this.sessionTokenUsage.outputTokens += usage.outputTokens;
+          this.addSessionUsage(usage);
           session.addTokenUsage(usage);
           if (showTokens && !tokensShown) {
             this.renderer.renderUsage(usage, this.sessionTokenUsage);
@@ -4637,8 +4698,7 @@ Session '${this.resumeSessionId}' not found.
         session.addMessage({ role: "assistant", content: response.content, timestamp: /* @__PURE__ */ new Date() });
         this.events.emit("message.after", { content: response.content });
         if (response.usage) {
-          this.sessionTokenUsage.inputTokens += response.usage.inputTokens;
-          this.sessionTokenUsage.outputTokens += response.usage.outputTokens;
+          this.addSessionUsage(response.usage);
           session.addTokenUsage(response.usage);
           if (this.shouldShowTokens()) {
             this.renderer.renderUsage(response.usage, this.sessionTokenUsage);
@@ -4773,7 +4833,11 @@ Session '${this.resumeSessionId}' not found.
     const apiMessages = [...messages];
     const extraMessages = [];
     const maxToolRounds = this.maxToolRoundsOverride ?? this.config.get("maxToolRounds") ?? DEFAULT_MAX_TOOL_ROUNDS;
+    const autoPauseIntervalRaw = this.config.get("autoPauseInterval");
+    const autoPauseInterval = typeof autoPauseIntervalRaw === "number" ? autoPauseIntervalRaw : DEFAULT_AUTO_PAUSE_INTERVAL;
     const baseSystemPrompt = (this.buildCurrentSystemPrompt() ?? "") + TOOL_CALL_REMINDER;
+    const pauseHint = autoPauseInterval > 0 ? `
+- Every ${autoPauseInterval} rounds the user will be asked whether to continue \u2014 use this as a natural checkpoint to report progress.` : "";
     const roundBudgetHint = this.planMode ? `
 
 [Tool Round Budget \u2014 Plan Mode]
@@ -4783,21 +4847,19 @@ You have a maximum of ${maxToolRounds} tool call rounds. You are in READ-ONLY Pl
 - Do NOT write shell commands or code blocks as a substitute for tool calls
 - Do NOT read the same file more than once
 - Call write_todos ONCE to present your plan, then give a text summary
-- If the user asks you to execute anything, respond: "Please type /plan execute to switch to execute mode."
-- Every ${AUTO_PAUSE_INTERVAL} rounds the user will be asked whether to continue.` : `
+- If the user asks you to execute anything, respond: "Please type /plan execute to switch to execute mode."${pauseHint}` : `
 
 [Tool Round Budget]
 You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan efficiently:
 - Prefer batch operations (e.g. global find-and-replace) over repetitive single edits.
 - Do NOT read the same file more than once \u2014 use the content from previous reads.
 - Prioritize the most critical tasks first in case rounds run out.
-- When remaining rounds are low, focus on completing the current task and summarizing.
-- Every ${AUTO_PAUSE_INTERVAL} rounds the user will be asked whether to continue \u2014 use this as a natural checkpoint to report progress.`;
+- When remaining rounds are low, focus on completing the current task and summarizing.${pauseHint}`;
     const systemPrompt = baseSystemPrompt + roundBudgetHint;
     const modelParams = this.getModelParams();
     const useStreaming = this.config.get("ui").streaming;
     const spinner = this.renderer.showSpinner("Thinking...");
-    const roundUsage = { inputTokens: 0, outputTokens: 0 };
+    const roundUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
     const supportsStreamingTools = useStreaming && typeof provider.chatWithToolsStream === "function";
     let consecutiveFreeRounds = 0;
     let lastToolCallSignature = "";
@@ -4904,6 +4966,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
         if (result.usage) {
           roundUsage.inputTokens += result.usage.inputTokens;
           roundUsage.outputTokens += result.usage.outputTokens;
+          roundUsage.cacheCreationTokens += result.usage.cacheCreationTokens ?? 0;
+          roundUsage.cacheReadTokens += result.usage.cacheReadTokens ?? 0;
         }
         if ("content" in result) {
           const hasWriteTools = toolDefs.some((t) => t.name === "write_file" || t.name === "edit_file");
@@ -4954,8 +5018,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
           });
           this.events.emit("message.after", { content: finalContent });
           if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-            this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-            this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+            this.addSessionUsage(roundUsage);
             session.addTokenUsage(roundUsage);
             if (this.shouldShowTokens()) {
               this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -4993,6 +5056,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
               if (genUsage) {
                 roundUsage.inputTokens += genUsage.inputTokens;
                 roundUsage.outputTokens += genUsage.outputTokens;
+                roundUsage.cacheCreationTokens += genUsage.cacheCreationTokens ?? 0;
+                roundUsage.cacheReadTokens += genUsage.cacheReadTokens ?? 0;
               }
               session.addMessage({ role: "assistant", content: genContent, timestamp: /* @__PURE__ */ new Date() });
               this.events.emit("message.after", { content: genContent });
@@ -5007,8 +5072,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
               const newMsgs2 = provider.buildToolResultMessages(result.toolCalls, syntheticResults, reasoningContent2);
               extraMessages.push(...newMsgs2);
               if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-                this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-                this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+                this.addSessionUsage(roundUsage);
                 session.addTokenUsage(roundUsage);
                 if (teeShowTokens && !teeTokShown) {
                   this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5116,12 +5180,12 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
         }
         const effectiveRound = round + 1;
         const remaining = maxToolRounds - effectiveRound;
-        if (AUTO_PAUSE_INTERVAL > 0 && effectiveRound > 0 && effectiveRound % AUTO_PAUSE_INTERVAL === 0 && remaining > 0) {
+        if (autoPauseInterval > 0 && effectiveRound > 0 && effectiveRound % autoPauseInterval === 0 && remaining > 0) {
           spinner.stop();
           process.stdout.write("\n");
           process.stdout.write(theme.warning(`\u23F8  Auto-pause: ${effectiveRound}/${maxToolRounds} rounds used, ${remaining} remaining
 `));
-          const recentHistory = roundToolHistory.slice(-AUTO_PAUSE_INTERVAL);
+          const recentHistory = roundToolHistory.slice(-autoPauseInterval);
           if (recentHistory.length > 0) {
             const toolCounts = /* @__PURE__ */ new Map();
             for (const rh of recentHistory) {
@@ -5204,6 +5268,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
           if (summaryResult.usage) {
             roundUsage.inputTokens += summaryResult.usage.inputTokens;
             roundUsage.outputTokens += summaryResult.usage.outputTokens;
+            roundUsage.cacheCreationTokens += summaryResult.usage.cacheCreationTokens ?? 0;
+            roundUsage.cacheReadTokens += summaryResult.usage.cacheReadTokens ?? 0;
           }
         } else {
           this.renderer.renderError(
@@ -5218,8 +5284,7 @@ Tip: You can continue the conversation by asking the AI to proceed.`
         );
       }
       if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-        this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-        this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+        this.addSessionUsage(roundUsage);
         session.addTokenUsage(roundUsage);
         if (this.shouldShowTokens()) {
           this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5316,7 +5381,12 @@ Tip: You can continue the conversation by asking the AI to proceed.`
       },
       getSessionTokenUsage: () => ({ ...this.sessionTokenUsage }),
       resetSessionTokenUsage: () => {
-        this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+        this.sessionTokenUsage = {
+          inputTokens: 0,
+          outputTokens: 0,
+          cacheCreationTokens: 0,
+          cacheReadTokens: 0
+        };
       },
       getGitBranch: () => this.gitBranch,
       getLastResponse: () => lastResponseStore.content,
@@ -5493,7 +5563,7 @@ program.command("web").description("Start Web UI server with browser-based chat
     console.error("Error: Invalid port number. Must be between 1 and 65535.");
     process.exit(1);
   }
-  const { startWebServer } = await import("./server-BQHIMEBH.js");
+  const { startWebServer } = await import("./server-MOTFJN6L.js");
   await startWebServer({ port, host: options.host });
 });
 program.command("user [action] [username]").description("Manage Web UI users (list | create <name> | delete <name> | reset-password <name> | migrate <name>)").action(async (action, username) => {
@@ -5726,7 +5796,7 @@ program.command("hub [topic]").description("Start multi-agent hub (discuss / bra
     }),
     config.get("customProviders")
   );
-  const { startHub } = await import("./hub-6V54V4O3.js");
+  const { startHub } = await import("./hub-HQ5QX7CV.js");
   await startHub(
     {
       topic: topic ?? "",

package/dist/{run-tests-6G65OGSL.js → run-tests-FRHDE3VB.js}

@@ -1,7 +1,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-FOFQAEU6.js";
+} from "./chunk-REWBXK2G.js";
 export {
   executeTests,
   runTestsTool

package/dist/{run-tests-P53FNUJY.js → run-tests-GISOOQZC.js}

@@ -2,7 +2,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-NP5KZVP6.js";
+} from "./chunk-7FOGK5TM.js";
 export {
   executeTests,
   runTestsTool